531Tzhan

7/30/2019 531Tzhan

1/30

Analysis of 50-year wind

data of the southernBaltic Sea for modelling

coastal morphological

evolution a case study

from the Darss-Zingst

Peninsula

OCEANOLOGIA, 53 (1-TI), 2011.pp. 489 518.

C 2011, by Institute of

Oceanology PAS.

KEYWORDS

Representative wind seriesStatistical analysis

Morphodynamic modelSouthern Baltic Sea

Wenyan Zhang1,

Jan Harff2

Ralf Schneider1

1 Institute of Physics,Ernst-Moritz-Arndt University of Greifswald,Felix-Hausdorff-Str. 6, Greifswald 17489, Germany;

e-mail: [email protected]

corresponding author

2 Institute of Marine and Coastal Sciences,Szczecin University,Adama Mickiewicza 18, Szczecin 70383, Poland

Received 6 October 2010, revised 1 March 2011, accepted 2 March 2011.

Abstract

High-resolution wind series in the southern Baltic Sea for the period of 19582007are analysed to generate representative climate input conditions for a multi-scalemorphodynamic model to simulate decadal-to-centennial coastline change. Fourseasonal wind classes, each characterized by a predominant distribution of winddirection and speed, are derived from statistical analysis. Further calibration of thisstatistical description is done by sensitivity studies of the model to generate similarcoastline changes of the Darss-Zingst peninsula as the measured data for the last

century. The coastline change of this area is then projected for the next 300 yearsbased on four different climate scenarios, through which impacts of accelerated sealevel rise and storm frequency on the long-term coastline change are quantified.

1. Introduction

Following the onset of the Littorina transgression (approximately 8000 calBP), the sea level in the southern Baltic Sea has reached a relatively stable

The complete text of the paper is available at http://www.iopan.gda.pl/oceanologia/

7/30/2019 531Tzhan

2/30

490 W. Zhang, J. Harff, R. Schneider

level with minor fluctuations in the range of only a few metres in the last6000 years (Kliewe 1995, Schumacher & Bayerl 1999). The rate of sea levelchange (in this study, the term sea level change refers only to eustaticchange) has generally been between 1 mm year1 and 1 mm year1 inthe southern Baltic Sea in the last 4000 years according to the results ofLampe (2005), which is of the same order of magnitude as the neotectonicmovements in this area (Harff et al. 2007). Along with the stable sealevel and neotectonic conditions, other processes such as climate change,hydrodynamics and sediment transport have become increasingly importantfor coastline evolution (Schwarzer & Diesing 2003).

In contrast to other waters, the Baltic Sea is distinguished by its greatvariety of coastal types. In general, till material predominates along thesouthern and south-eastern coasts, while hard-bottom and rocky shores aretypical on northern coasts (Schiewer 2008). The Baltic Sea can be describedas a tideless semi-enclosed marginal sea. The hydrodynamics of the Balticis characterized mainly by complicated meso-to-large scale wind-drivencurrents and local-scale wind-induced waves. Tides coming from the NorthSea attenuate quickly after entering the Baltic Sea through a series of narrowchannels. The tidal range in the southern Baltic area is no more than 15 cm,while large-scale meteorological situations can excite a storm surge with

11.5o 12.0 12.5 13.0 13.5 14.0o o o o o

55.2

55.0

54.8

54.6

54.4

54.2

54.0

o

o

o

o

o

o

o

longitudeE

latitu

de

N

Figure 1. General features and location of the Darss-Zingst peninsula. Hindcastedwind series at five points (triangles) are analysed to generate the representativewind inputs for the long-term model

7/30/2019 531Tzhan

3/30

Analysis of 50-year wind data of the southern Baltic Sea . . . 491

water level changes of the order of 1.5 m within one day. The Darss-Zingst

peninsula (Figure 1) on the southern Baltic was formed after the postglacial

transgression (Schumacher 2002, Lampe 2002) and is composed of two mainparts. The exterior part is a triangularly shaped barrier island with two

wings extending south-westwards (Fischland-Darss) and eastwards (Darss-

Zingst), and a headland (Darsser Ort) linking the two wings in the north.The formation of the barrier island is the result of a combination of climate

change, hydrodynamics and sediment transport, which still remains active

today. The interior part consists of a chain of lagoons (the Darss-ZingstBodden), which are subject to progressive phytogenic silting-up. Thewesterly exposed coast of Darss and the northerly exposed coast of Zingst

are characterized by strong abrasion of the cliff coast and the flat beach

coast, as well as a rapid accumulation at the top of the headland (DarsserOrt) as a result of the abundant sediment supply brought by the wind-

induced longshore currents. The eastern extension of the peninsula is theBock sand flat, which is separated from the southernmost tip of Hiddensee

Island by a dredged channel. Bock Island is like a container, where sedimenttransported southwards along Hiddensee and eastwards along the Zingst

coast accumulates. The particular evolution of the Darss-Zingst peninsula

may serve as a good example to study coastal evolution under long-termclimate change, and has instigated several descriptive and conceptual studiesin the last 100 years (Otto 1913, Kolp 1978, Lampe 2002, Schumacher 2002).

In contrast to traditional geological and sedimentological studies based

on field observation and analysis, morphodynamic modelling of coastalevolution based on process concepts is in its infancy, owing to its dependence

on computer power, which has only recently become available. Process-based models can be divided into three categories according to their object

of study on different time scales: (1) real-time simulation on time scalesfrom tidal to seasonal periods, (2) medium long-term simulation on time

scales from annual, decadal to centennial and millennial periods, and (3)

extreme long-term or geological time scale (longer than 10 000 years scale)simulation. Models for the first and third category are well developed todayand a wide range of such models is available. However, the development of

models for the second category (hereafter referred to as long-term model)

has yet to reach maturity (Fagherazzi & Overeem 2007). A common wayof simulating decadal-to-centennial coastal morphological evolution is to

extrapolate the real-time calculation (the first type of model) to longer time

periods. However, poor results are usually obtained when these modelsare applied directly on such a time scale. This is due mainly to three

reasons: (1) the time step of calculation in high-resolution process-based

models (the first type of model) is determined by the shortest time-scale

7/30/2019 531Tzhan

4/30

7/30/2019 531Tzhan

5/30


construction and its modeling strategy for a long-term scale is introducedin Zhang et al. (2010). In this paper we introduce mainly the conceptsof representative climate input conditions for the morphodynamic modeland the methodology for generating representative climate input conditions.The coastline changes of this area in the next 300 years, based on fourdifferent climate scenarios derived from different studies, are then projectedby the model, through which the impacts of accelerated sea level rise andstorm frequency on long-term coastline change are quantified.

2. Material and methods

2.1. Brief description of the model

Simulation of the decadal-to-centennial morphological evolution of theDarss-Zingst peninsula is based on a multi-scale morphodynamic modelconsisting of 8 modules to calculate different physical processes that drivethe evolution of the specific coastal environment. The two-dimensionalvertically integrated circulation module, the wave module, the bottomboundary layer module, the sediment transport module, the cliff erosionmodule and the nearshore storm module are real-time calculation modulesthat aim to solve short-term processes. A bathymetry update moduleand a long-term control function set, in which the reduction conceptsand technique for morphological update acceleration are implemented, areintegrated to up-scale the effects of short-term processes to a decadal-to-

centennial scale.

2.2. Wind analysis

Boundary input conditions for a long-term (decadal-to-centennial) mor-phodynamic model such as time series of tides, winds, waves and massflux cannot be specified at a centennial time span owing to the lack ofmeasurements. On the other hand, even if detailed, measured time seriesof boundary conditions were provided, it would be an extremely time-consuming job for a high-resolution process-based model to calculate thecentennial-scale coastal evolution with the measured time series. This isbecause the time step of calculation in high-resolution process-based models

is determined by the shortest time scale process, which usually has tobe solved on a time scale of seconds or minutes. Representative inputconditions, which are generated by the statistical analysis of the measuredtime series, provide an effective way of solving the input problem for thelong-term model. The generation of representative input conditions is basedon the concept of input reduction for long-term modelling (de Vriend et al.1993a,b). The criterion for judging the validity of the representative input

7/30/2019 531Tzhan

6/30

7/30/2019 531Tzhan

7/30


Oc to be r N ov emb er D ec emb er January February

March April May

June July August

September

N

E

S

W

October,November,December,January,Februaryaredominatedby WSW winds

March, April,MayaredominatedbyE-W winds

June,July, Augustaredominatedby WNW winds

Septemberisdominatedby W winds

30%

20%

10%

Figure 2. Four wind classes, each with a predominant wind direction, are classifiedfrom the wind roses of 12 months based on the data from 1958 to 2007

results indicate that the wind time series at Point 3 is closest to the meanvalue of the series at the five points (with a value of 0.34 m s1 for the RMSEof component u and 0.22 m s1 for the RMSE of component v). Statisticalresults indicate that the southern Baltic Sea is dominated by westerly windsand the 50 year-averaged wind speed is 7.5 m s1 in the Darss-Zingst area.The ratio of westerly winds (hours) to easterly winds (hours) is about 18:11.The distribution of wind directions of each month in this period showsthat the winds in the Darss-Zingst area can be classified into four seasonalclasses (Figure 2). Each class has a predominant distribution of winddirection. By combining the monthly average wind speed profiles, Class 1(October, November, December, January and February) can be identifiedas a winter class with relatively strong wind conditions; the prevailing winddirection is WSW. Class 3 (June, July and August) can be identified asa summer class with mild wind conditions dominated by the WNW winds.Class 2 (March, April and May) and Class 4 (September) are transitionalclasses with moderate wind conditions. Class 2 is dominated by the East-West balanced winds and Class 4 is dominated by westerly winds. TheWeibull distribution is utilized to analyse the wind strength. The Weibulldistribution function is a two-parameter function, for wind speed

f( 0; k, ) = k

k

k1exp

k, (2)

where f is the probability density function (PDF) of the wind speed, represents the wind speed, k > 0 is the shape parameter, and > 0 is the

7/30/2019 531Tzhan

8/30


scale parameter. The cumulative distribution function is given by

F() = 1 exp

k

. (3)

With a double logarithmic transformation, eq. (3) can be written as

ln{ ln[1 F()]} = k ln k ln . (4)Knowing F() and from the wind speed data, the value of k and canbe determined by least squares fitting using eq. (4).

Table 2. Monthly long-term shape parameters k, scale parameters and PearsonsChi-square 2 of the Weibull distribution at Point 3 for the period 19582007

Month Shape (k) Scale () 2

with Point3 [m s1] with Point3 [m s1]

Jan. 2.44 9.40 0.015Feb. 2.41 8.97 0.015March 2.29 8.52 0.018April 2.24 7.16 0.021May 2.22 6.70 0.019June 2.17 6.66 0.022July 2.12 6.84 0.021Aug. 2.20 6.87 0.019Sept. 2.27 7.61 0.023Oct. 2.40 8.89 0.017

Nov. 2.51 9.29 0.016Dec. 2.52 9.37 0.011

The Weibull parameters for each month (Table 2) are obtained byapplying eq. (4) to the 50-year wind series. Pearsons Chi-square test isused to evaluate the performance of the Weibull fitting, which is given by

2 =Ni=1

(Oi Ei)2/Ei, (5)

where Oi is the measured frequency for bin i (the wind speed data is dividedinto 60 bins at intervals of 0.5 m s1), and Ei is the expected frequency for

bin i, which is calculated by

Ei = k(F(i/2) F(i/2 0.5)), (6)where k is the size of the wind speed series, and F is the cumulativedistribution function given by eq. (3).

Results of Pearsons Chi-square test show satisfactory fitting of theWeibull distribution to the wind data (Table 2). Weibull parameters for

7/30/2019 531Tzhan

9/30


0 50 100 150 2 00 250 3 00 3500

5

10

direction[degree]

Class1

0 5 10 15 20 25 300

5

10

15

rep.forClass1rep.forClass2rep.forClass3

k=2.1532

l =6.8

frequency

[%]

0 50 100 150 2 00 250 300 3500

5

10

direction[degree]

Class3

0 50 100 150 2 00 250 3 00 3500

5

10

direction[degree]

Class2

frequency

[%]

0 50 100 150 2 00 250 300 3500

5

10

direction[degree]

Class4

a

b

f(x

)[%]

x[ms ]-1

k=2.4151

l =9.08

k=2.1635

l =7.435

Figure 3. (a) Distribution of wind directions for each class: 90 indicates aneasterly wind and 270 a westerly wind; (b) the Weibull distribution of wind speedsfor each class

the months in Class 1 indicate their similar distributions of wind strength.The months in Class 3 also have similar Weibull parameters. The Weibullparameters of the three months in Class 2 indicate a decreasing trend of wind

strength. The average term of the wind strength of this class is reflectedin the April distribution. Based on the similarities of the monthly Weibullparameters within the same class, the Weibull distribution for each class is

obtained by applying eq. (4) to the wind series of the months within thesame class. The results are shown in Figure 3b (parameters of Class 4 are notshown as they are already listed in Table 2). The concept of representativemonthly wind series is introduced in this study. A representative monthlywind series is composed of 720 (hours in a month) synthetic wind elements.This is able to reflect statistically the features (spectrum) of a wind class,and thus represents the months of one class. The use of representative

7/30/2019 531Tzhan

10/30

7/30/2019 531Tzhan

11/30


contains a long series of wind elements with similar wind directions willinduce larger wind fetch and stronger waves in the model calculation).As the research area is dominated by westerly winds, the division ofthe westerly wind sub-groups (SWW and WNW) is analysed. Nodivision is carried out in the other groups (NNE, NEE, ESE, SES, SSW and NWN). The resultant representative wind series forclass 1 after step (2) and the division scheme of the westerly windsub-groups are shown in Figure 5. Sensitivity experiments (whichwill be described in the following part) are carried out to obtain theoptimum division of the westerly wind sub-groups capable of inducingapproximate coastline change in the morphodynamic model comparedto the measured data;

Class1

a(SW-W)sub-group1

2 3 4

5

67

8

0 100 200 300 400 500 600 700

a(W-NW)sub-group

timeseries[h]

15ms-1 E

S

W

N

Figure 5. Top panel: Generated representative wind series for Class 1 after step2. Middle panel: Division of the westerly wind sub-groups by a factor of two inrepresentative wind series 1. Bottom panel: Division of the westerly wind sub-groups by a factor of four in representative wind series 1

(4) When the optimum division of wind sub-groups is obtained throughthe sensitivity studies, the sub-groups of each representative windseries are arranged according to the short-term cyclical term on anhourly or daily scale, and the resultant representative series need to

be further corrected by longer-term (yearly) trend and cyclical terms.Longer time scales of trend and cyclical terms that exceed the timespan of the hindcast data (50 years) also need to be analysed accordingto the studies of long-term climate change. In our present study,the trend and cyclical term at scales longer than 50 years for thehindcast and projection of the 300 years evolution of the Darss-Zingstpeninsula are neglected and the normal wind conditions that exclude

7/30/2019 531Tzhan

12/30


storm information are assumed to be periodic every 50 years. Thisassumption is based on the analysis of the trends and cycles of the50-year wind data, which is introduced as follows.

The yearly trend term of wind series is given by

{Vt+n} = {Vt} + mL,n, (7)where {Vt+n} represents the average value of wind series in the nth timeperiod (year), {Vt} represents the average value of the first wind series(year), mL,n is the yearly trend term of the wind series. To determine theyearly trend terms, the Weibull parameters for each year from 1958 to 2007are calculated. The linear best-fit functions of the parameters indicate veryslight trend terms. For the scale parameter , the best-fit function is

y = 0.0004 0.03, (8)where y denotes the value of , and denotes the year (from 1958 to 2007).For the shape parameter k, the best-fit function is

y = 0.0004 + 1.447. (9)

These trend terms are too small to be considered on a decadal-to-centennialscale, so the yearly-scale trend term of wind strength is assumed to be zero.

The cyclical term of wind series can be divided into a long-term (yearly)cyclical term (SL,n) and a short-term (hourly to daily) cyclical term(SL,h). The short-term cyclical term is obtained by calculating the auto-

correlation coefficients of hourly wind speed and wind direction series withtime lags from 1 hour to 8760 hours. Results show that the value of theautocorrelation coefficient decreases abruptly from 0.95 to 0.1 in the first72 hours in both series, and is maintained in the range from 0.1 to 0.1in lags from 72 to 8760 hours. The loss of correlation within a short timein both series indicates that there are no short-term cyclical terms in thewind series. The yearly cyclical term is shown in both class-averaged windspeed series and wind direction series, which indicates similarities of windseries within each class on a yearly scale. Based on the results of the cyclicalterms, each representative wind series can be regarded as an independentseries not correlated with the others.

With the information on trend and cyclical terms to hand, we canconclude a modelling strategy that the generated representative monthlywind series for each class, which serve as climate inputs for the model, ismerely repeated in every cycle of model calculation (each cycle calculatesone years morphological change) without any trend correction. The samerepresentative wind series are used in the hindcast of the last 300 years aswell as the forward projection to the next 300 years. The use of the same

7/30/2019 531Tzhan

13/30


wind input conditions in the future projection is based on the IPCC (2007),which indicates that there are no consistent agreements on the future changeof average or extreme wind speeds in Europe.

2.2.2. Storm analysis

Most information about extreme wind events is filtered in the generationof representative wind series, as extreme wind events make up only a smallpercentage of the whole time period. The statistics of hindcast wind datafrom 1958 to 2007 indicate that extreme wind events are frequent in thesouthern Baltic area and may play an important role in reshaping thecoastline of the Darss-Zingst peninsula. Normally, the definition of a stormis related to water level variation and wind speeds. However, as only thewind data for this 50-year period are available, we here define a storm asa continuous time series with maximum wind speeds greater than 20 m s1,minimum wind speeds greater than 14 m s1 and of duration longer than24 hours. According to this criterion, 57 storms occurred in the Darss-Zingst area during 19582007, 8 of which were from the east and the restfrom the west. Statistical results indicate that January and November canbe described as storm months: 31 storms took place in this period. Thedistribution of storm directions indicates that WNW is the most probabledirection for a storm in this area, with a percentage of 43%. The mostprobable direction for an easterly storm is NE, with a percentage of 65%in the 8 storms. The annual maximum wind speed profile indicates thatstorms occur almost every year and that there is no distinct trend in thevariation of the storm strength in this 50-year period.

Extreme value theory is applied to analyse the storms. The Gumbeldistribution is used to calculate the return period of storms. The probabilitydensity function of the Gumbel distribution is

f() =1

exp

exp

, (10)

where is the scale parameter, is the location parameter, and is themaximum wind speed of the year.

The cumulative distribution function of the Gumbel distribution is givenby

F() = exp exp

. (11)

Following a double logarithmic transformation, eq. (11) can be written as

ln[ ln F()] =

. (12)

Knowing F() and from the statistics of the annual maximum windspeed, the values of and can be obtained by least squares fitting using

7/30/2019 531Tzhan

14/30


17 18 19 20 21 22 23 24 2526 27 2829 300.01

0.10

1.00

annualmaximumwindspeed[x, ]ms-10 20 40 60 80 100

21

22

23

24

25

2627

28

29

returnperiod[yrs]

max

imumw

ind

speed

[ms

]-1

Best-fitGumbeldata

pro

bab

ilit

yo

fexceedance

:1

-F(x)

a b

Figure 6. (a) Probability of the annual maximum wind speed being exceeded; (b)return periods of maximum wind speeds (b)

eq. (12). The best-fit Gumbel distribution of the annual maximum windspeed for the period of 19582007 is given by = 1.498 and = 20.49. Theresultant Gumbel fitting curve is shown in Figure 6a. The return period ofthe annual maximum wind speed is thus given by

R() =1

1 F() . (13)

The curve of R() is shown in Figure 6b. Some maximum wind speedswith their return periods and probability of occurrence are listed in Table 3.

Table 3. Calculated return periods and probabilities of occurrence of annualmaximum wind speeds on the Darss-Zingst peninsula based on 50-year data

Annual maximum Return period Probabilitywind speed [m s1] [years] of occurrence [%]

28.6 100 127.4 50 1.826.2 25 1.825.8 20 3.524.5 10 723.1 5 17.521.5 2 4718.3 1 96.5

It is not realistic for the morphodynamic model to include every windstorm with a different return period. According to the distribution of windstorm directions and the frequency of the storms that blew in the researcharea in the period of 19582007, one annual storm from the WNW anda once-every-n-years storm from the NE are included in the model. Here,n is a variable (which should range between 5 and 10 according to the

7/30/2019 531Tzhan

15/30


statistical result) that results from correction of the wind-induced wavespectrum aiming at inducing similar coastline change to the measured data.The maximum wind speed is 21.5 m s1 in both storms and the duration is48 hours (wind speed above 14 m s1, according to the definition of a storm),which is the typical duration of a storm in the southern Baltic Sea accordingto the statistical results of wind storms in 19582007. The wind speed isassumed to increase linearly from 14 m s1 to 21.5 m s1 and then decreaselinearly to 14 m s1 within this time span.

2.3. Calibration of the representative wind series

Four primary representative wind series are generated according to themethodology presented in section 3. However, these are not yet the final

series serving for the model boundary input as the internal variation ofthese series such as the ordering of the wind sub-groups and the wind fetch(determined by the division of wind sub-groups) may significantly influencethe simulation results. In order to obtain a wind series that induces a similarcoastline change as the measured data (Figure 7), a series of model runs arecarried out to test the sensitivity of the simulation results to the variationof the representative wind series.

cliffcoast

opening

coastalloss(mper100yr)-80

Figure 7. Measured coastline change of the Darss-Zingst peninsula in the 20thcentury (from the State Agency for Environment and Nature, Rostock (StAUN1994)). A negative number denotes coastline retreat, a positive number denotescoastline accretion (m per 100 years)

7/30/2019 531Tzhan

16/30

7/30/2019 531Tzhan

17/30

7/30/2019 531Tzhan

18/30


2.4. Validation of the representative wind series

A digital elevation model (DEM) of the research area for the year1696 was reconstructed on the basis of high-resolution bathymetric andtopographic data sets measured in modern times (Zhang et al. 2011).Based on the reconstructed DEM, a recent sediment map, an isostatic map,an eustatic scenario for the last three centuries (Meyer et al. 2008) andvalidated modules, the model was applied to hindcast the coastal evolutionof the Darss-Zingst peninsula from 1696 to 2000 without taking into accountanthropogenic influence (Zhang et al. 2011). The calibrated representativewind series serve as input conditions for the model. Successful validationof the representative wind series was shown by comparison between themodelled coastline change and the measured data along the peninsula with

a RMSE = 61 m (which is about 1/5 of the averaged coastline changefor the last 300 years). The simulated coastline in different time periodsindicates a smooth evolution of the area in the last 300 years. Most ofthe coastline has been retreating except two parts: (1) the headland andits eastern side and (2) Bock Island. These two areas act like reservoirswhere sediment converges, but the mechanisms driving their evolution aredifferent. The growth of the headland is a combination of long-term wavedynamics (wave breaking, longshore currents) and short-term storm effects.The development of Bock is a residual effect of long-term wave dynamics(Zhang et al. 2010). The simulated results agree with other studies onthe Darss-Zingst peninsula (Lampe 2002, Milbradt & Lehfeld 2002, Froehle

& Dimke 2008).

3. Results

Based on a successful validation, the model is used to project themorphological evolution of the Darss-Zingst peninsula during the next300 years without consideration of any coastal protection measures. Theeffects of sea level rise and storm frequency on coastline change in thesouthern Baltic are quantified. Four different climate scenarios are designed,based on existing studies of climate change in the southern Baltic Sea oradjacent area (North Sea). All scenario runs use the same representative

wind series described in section 3.1. The differences among these runsare the parameterization of the storm frequency and the rate of sea levelchange. The first scenario (Scenario 1) assumes an average sea level rise of2 mm year1 (Meyer et al. 2008). The storm frequency in this run remainsthe same as the 50-year statistical results (i.e. an annual WNW storm anda once-every-5-years NE storm). Though there is little consistent evidenceamong different studies that shows changes in the projected frequency of

7/30/2019 531Tzhan

19/30


extreme wind events at either a global or a regional scale (IPCC 2007), inorder to quantify the effects of storms on the coastline change, an increase ofthe storm frequency by 20% (both for storms from the WNW and the NE)compared to the 50-year results is assumed in the second climate scenario(Scenario 2). A sea level rise of 2 mm year1 is also parameterized in thesecond scenario. The third climate scenario (Scenario 3) assumes an averagesea level rise of 3 mm year1 according to the projection results (19902100)of the sea levels of the Baltic Sea described in Meier et al. (2004). The stormfrequency remains the same as the 50-year statistical results in the thirdscenario. In the fourth climate scenario (Scenario 4) both the rate of sealevel rise and storm frequency are increased (i.e. a 3 mm year1 sea levelrise and an increase in storm frequency by 20% compared to the 50-yeardata).

3.1. Morphological change

The coastline change in most parts of the peninsula is acceleratedcompared to the change in the last 300 years owing to the sea level risein Scenario 1 (Figure 9). An increment of 1015 m per 100 years in thecoastline retreat on the Darss coast is anticipated compared to the rates of

Figure 9. Projected coastal evolution of the Darss-Zingst peninsula in the next300 years based on climate Scenario 1. The red circles indicate the active channels

7/30/2019 531Tzhan

20/30

7/30/2019 531Tzhan

21/30

7/30/2019 531Tzhan

22/30


The projected headland in Scenario 3 is much narrower than in Scenarios 1and 2, even though it is still growing. An increased retreat of about 150 min the western part and about 165 m in the eastern part of the headland(compared to Scenario 2) is projected in Scenario 3. The thinning of theheadland is caused mainly by the effects of accelerated sea level rise. Asthe sea level rises more rapidly than usual (Scenario 1), sediments (eitherfrom coastal erosion or supplied by longshore currents from other sources)become easier to trap in the deeper offshore area, where waves are difficult toreach (this can be identified from the profile changes shown in Figure 11).Trapping of sediment in the offshore area reduces the number of sourcesfor the headlands growth; on the other hand the rise of sea level furthercounterbalances the sedimentation around the headland. With a smallersediment supply, the headland thus becomes ever narrower. The acceleratedsea level rise is not only responsible for the thinning of the headland, butalso causes significant changes in the Zingst area: this is mostly submergedby water in Scenario 3, leaving only several discrete sand flats. Hiddenseesuffers a similar fate and is split into two main islands. Five new channelsare formed in Scenario 3, two of which are on Darss, two are on Zingst andone is in the Hiddensee area. The results of Scenario 3 also indicate thatthe Zingst coast is most sensitive to the accelerated sea level rise.

The projected coastline in Scenario 4 seems quite similar to Scenario 3,with minor differences (e.g. an average increased coastline retreat of 30 mon Darss compared to Scenario 3) in most parts. The largest difference ofthe coastline between these two scenarios lies in the headland. The headlandprojected by Scenario 4 becomes broader than in Scenario 3: this is due tothe increased storm frequency, which provides additional sediment sourcesfor the headland, even though a large part of the sediment is trapped in theoffshore area. Another difference between Scenarios 3 and 4 is the offshorearea. Scenario 4 induces more sedimentation in the offshore area as a resultof the increased storm frequency. This is especially evident in the Zingstarea, where the 5 m and 7.5 m isobaths extend about 190 m and 110 mnorthwards respectively.

3.2. Coastal profile change

A plot of the profiles perpendicular to the coastline can help toshow more details of the cross-shore changes induced by different climatescenarios. In Figure 11, changes of the profile located on the Darss coastare compared. The horizontal resolution of the profiles is about 100 m inthe coastline area (between 100 m and 500 m in the cross-shore directionshown in Figure 11) and gradually decreases to 300 m at the 13 m isobath.Resolution of the further offshore area is about 400 m. Projection results

7/30/2019 531Tzhan

23/30


indicate remarkable profile changes in the nearshore and offshore areas. Allfour scenarios anticipate erosion in the nearshore area (where the waterdepth is less than 3 m) and deposition in the adjacent offshore area.A longshore bar develops as a result of sedimentation in the offshore area.The position of the longshore bar is not always fixed: it moves upwards asthe sea level rises, along with further development. The difference betweenthe profiles in Scenarios 1 and 3 is remarkable, even though the sea levelrise increases by just 30 cm in the latter. This is because the accelerated sealevel rise enables waves to act further up the beach profile and cause strongererosion. The increased storm frequency also causes significant changes onthe profile. Scenario 2 produces a quite similar profile to Scenario 3. Thesimilarity of profile change between these two scenarios indicates that anaccelerated sea level rise of 3 mm year1 and a 20% increase in stormfrequency have almost the same effects on the coastline change of Darss.However, the combination of these two factors in Scenario 4 does not causea linear effect in which the individual effects of these two factors on theprofile can simply be summed. Comparison of Scenario 4 with the othertwo scenarios (Scenario 2 and 3) indicates that the profile evolves into analmost equilibrium state in these scenarios.

4. Discussion

4.1. Long-term modelling strategy

A long-term morphological simulation cannot take into account all theprocesses involved, especially those stochastic processes taking place ona short time scale (e.g. one heavy precipitation event) owing to a lack ofdata and practical run-time limits. In order to solve the problem of modelinput, concepts of input reduction are implemented in our modelling work.Input reduction refers to the filtering of the climate input conditions fora long-term model. Representative climate time series, which are generatedby statistical analysis of the measured data and corrected by sensitivitystudies, serve as input for the long-term model. A critical criterion forevaluating the reliability of the representative climate time series is whetherthe model computation with the representative input conditions produces

similar results to the reference data. Thus, calibration and validation of therepresentative time series are very important before the final applicationof the model. Calibration of the representative wind series in this workis based on a series of sensitivity studies in which the effects of stormfrequency, wind fetch and ordering of wind sub-groups on the coastlinechange are quantified. The representative wind series are validated bycomparison between the model results and measured coastline change in

7/30/2019 531Tzhan

24/30


the last 300 years. Hindcast results indicate long-term wave dynamics(wave breaking, longshore currents) and short-term storms as two dominantfactors influencing the coastline change of the Darss-Zingst peninsula in thelast three centuries. Compared to these two factors, long-term sea levelchange played a minor role in driving coastal evolution in this time periodbecause of its relatively low rate, which is about 1 mm year1 according toHupfer et al. (eds.) (2003) and Ekman (2009).

Morphodynamic evolution of the Darss-Zingst coastline is significantlyinfluenced by regional climate factors such as sea level change and winds.To represent the climate factors as input for the long-term model, hourlywind series for a 50-year period (19582008) are analysed to generate therepresentative time series; waves and currents are calculated in the modelusing the wind field as input; the long-term (centennial) sea level changeis parameterized on the basis of existing climate studies (Voss et al. 1997,Meier et al. 2004, IPCC 2007). Although satisfactory hindcast results areproduced, the influences of other factors such as coastal engineering workon coastline change need to be carefully evaluated when the model is usedfor future projections. Coastal engineering work has provided additionalprotection for the Darss-Zingst coastline since the last century (Froehle& Kohlhase 2004) and will continue to do so for the foreseeable future.Such anthropogenic influence is excluded in our model, as projection resultswithout anthropogenic influence should help to make coastal managementmore efficient by indicating how nature acts on coastline change, thusproviding useful information for coastal engineering work.

4.2. Storm effects on long-term coastline change

Our model results suggest that both accelerated sea level rise andincreased storm frequency have significant effects on the coastline erosionof the Darss-Zingst peninsula on a centennial scale. The effect of sealevel rise on long-term coastline change is commonly recognized in currentstudies. However, the effects of storms on the long-term coastline changeindicated in our model results seem to be inconsistent with some otherstudies (Douglas & Crowell 2000, Zhang et al. 2002) in which storm erosionof the beach was found to be episodic but not secular, as the beach profile

recovers after each storm. Actually, the importance of storm effects onthe coastline change found in this study does not conflict with the studiesmentioned above, as there are many differences between the research areas.In a study on the Texas coast Morton et al. (1994) discovered four dominantprocesses in beach recovery: (1) rapid forebeach accretion under mildweather conditions, (2) backbeach aggradation, (3) dune formation and(4) dune expansion and vegetation recolonization. They also found that

7/30/2019 531Tzhan

25/30

7/30/2019 531Tzhan

26/30

7/30/2019 531Tzhan

27/30

7/30/2019 531Tzhan

28/30


Dissanayake D. M. P. K., Roelvink J. A., 2007, Process-based approach on tidalinlet evolution Part 1, Proc. 5th Symp. River, coastal and estuarine

morphodynamics, 225269.

Douglas B. C., Crowell M., 2000, Long-term shoreline position prediction and errorpropagation, J. Coast. Res., 16 (1), 145152.

Ekman M., 2009, The changing level of the Baltic Sea during 300 years: A clueto understanding the Earth, Summer Inst. Historical Geophys., Aland Islands,155 pp.

Fagherazzi S., Overeem I., 2007, Models of deltaic and inner continentalshelf landform evolution, Annu. Rev. Earth Pl. Sci., 35(1), 685715, doi:10.1146/annurev.earth.35.031306.140128.

Froehle P., Dimke S., 2008, Analysis of potential long shore sediment transportat the coast of Mecklenburg-Vorpommern, [in:] Proc. Chinese-German Joint

Symposium of Hydraulic and Ocean Engineering, U. Zanke et al. (eds.), Inst.Hydraulic and Water Resources Engineering (Hrsg.).

Froehle P., Kohlhase S., 2004, The role of coastal engineering in integrated coastalzone management, Coast. Rep. 2, 167173.

Harff J., Lemke W., Lampe R., Luth F., Lubke H., Meyer M., Tauber F., SchmolckeU., 2007, The Baltic Sea coast-A model of interrelations among geosphere,climate, and anthroposphere, [in:] Coastline changes: Interrelation of climateand geological processes, J. Harff, W. W. Hay & D. M. Tetzlaff (eds.), Geol.Soc. Am. (GSA) Spec. Papers 426, 133142

HELCOM, 2006, Climate change in the Baltic Sea area, HELCOM StakeholderConf. Baltic Sea action plan, HELCOM thematic assessment in 2006, HelsinkiCommiss., Helsinki.

Hupfer P., Harff J., Sterr H., Stigge H.-J. (eds.), 2003, Die Wasserstande an derOstseekuste. Entwicklung Sturmfluten Klimawandel, Archiv fur Forschungund Technik an der Nord- und Ostsee, Die Kuste H. 66, 331 pp., Boyens & Co.KG, Heide i. Holstein.

IPCC, 2007, Climate change 2007: The physical science basis, Contribution ofWorking Group I to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change, Cambridge Univ. Press, Cambridge.

Jimenez J. A., Arcilla A. S., 2004, A long-term (decadal scale) evolution model formicrotidal barrier systems, Coast. Eng., 51 (89), 749764.

Jones O. P., Petersen O.S., Hansen H. K., 2007, Modelling of complex coastalenvironments: Some considerations for best practise, Coast. Eng., 54 (10),717733.

Klavins M., Briede A., Rodinov V., 2009, Long term changes in ice and dischargeregime of rivers in the Baltic region in relation to climatic variability, Clim.Change, 95 (34), 485498.

Kliewe H., 1995, Zeit- und Klimamarken in Sedimenten der sudlichen Ostsee undihrer Vorpommerschen Boddenkuste, J. Coast. Res., Special Issue. 17, 181186.

Kolp O., 1978, Das Wachstum der Landspitze Darsser Ort, PetermannsGeographische Mitteilungen 122, 103111.

7/30/2019 531Tzhan

29/30


Lampe R., 2002, Holocene evolution and coastal dynamics of the Fischland DarssZingst peninsula, Greifswald Geographische Arbeiten, 27, D1, 155163.

Lampe R., 2005, Late-glacial and Holocene water-level variations along the NEGerman Baltic Sea coast: review and new results, Quatern. Int. Vol. 133134,121136.

Latteaux B., 1995, Techniques for long-term morphological simulation under tidalaction, Mar. Geol., 126 (14), 129141.

Meier H. E. M., Broman B., Kjellstrom, E., 2004, Simulated sea level in past andfuture climates of the Baltic Sea, Climate Res., 27 (1), 5975.

Meyer M., Harff J., Gogina M., Barthel A., 2008, Coastline changes of theDarss-Zingst peninsula a modeling approach, J. Marine Syst., 74 (Suppl. 1),

S147S154.

Milbradt P., Lehfeldt R., 2002, Littoral processes at micro-tidal coasts of thesouthern Baltic Sea, Adv. Hydro-Sci. Eng.

Morton R. A., Paine J.G., Gibeaut J. G., 1994, Stages and durations of post-storm beach recovery, southeasterly Texas coast, U.S.A., J. Coast. Res., 10 (4),

884908.

Otto T., 1913, Der Darss und Zingst: Ein Beitrag zur Entwicklungsgeschichte dervorpommerschen Kuste. 13., Jahresber. Geogr. Ges. Greifswald, 19111912,237485.

Roelvink J.A., 2006, Coastal morphodynamic evolution techniques, Coast. Eng.,53 (23), 277287.

StAUN State Agency for Environment and Nature (StAUN) Rostock,1994, Generalplan Kusten- und Hochwasserschutz Mecklenburg-Vorpommern,Landesentwicklung und Umwelt Mecklenburg-Vorpommern, Ministerium furBau, 108 pp.

Schiewer U., 2008, Chapter 2: The Baltic coastal zones, [in:] Ecology of Balticcoastal waters, Springer, doi: 10.1007/978-3-540-73524-3.

Schumacher W., 2002, Coastal evolution of the Darss Peninsula, GreifswalderGeographische Arbeiten, 27, D2, 165168.

Schumacher W., Bayerl K.-A., 1999, The shoreline displacement curve of RuegenIsland (Southern Baltic Sea), Quatern. Int., 56 (1), 107113, doi: 1016/S1040-6182(98)00027-5.

Schwarzer K., Diesing M., 2003, Coastline evolution at different time scales examples from the Pomeranian Bight, southern Baltic Sea, Mar. Geol.,

194 (12), 79101.

Trouet V., Esper J., Graham N. E., Baker A., Scourse J. D., Frank D. C., 2009,Persistent positive North Atlantic Oscillation mode dominated the MedievalClimate Anomaly, Science, 324 (50)2009, 78.

Voss R., Mikolajewicz U., Cubasch U., 1997, Langfristige Klimaanderungen durchden Anstieg der CO2-Konzentration in einem gekoppelten Atmosphare-Ozean-Modell, Ann. Meteorol., 34, 34.

7/30/2019 531Tzhan

30/30

531Tzhan

Documents

Transcript of 531Tzhan